It will increase the Blue component in the sky glow, but as it is Blue, it provides allmost no visible light

The overall light pollution will likely decrease because most of the time, the LED luminaires are of way lower output flux than the HPS they replace, so there is less light overall, including direct light and reflected from the Earth surface light

Well, with the blue component there is one aspect extra why that would be off a less problems: Most materials on ground tend to absorb the blue way more than the longer wavelengths. So there is way less reflected of it...

Dont think it applies all that well to artificial materials like concrete or other paving materials.. And definitely not to snow

A truely complex process involving different interactions in different weather situations.

In clear dry air where Rayleigh scattering (Scatter ∝ 1/λ⁴) is dominant in the air above a town or city, the bluer content of LED will cause more local light-pollition even if lower levels on the ground are used.

A different relation with larger pollutants than air molecules exists. Scattering here is proportional to the inverse wavelength (Scatter ∝ 1/λ). These conditions however are rare more readily occuring in the day with the sun's action. Chemical smogs, sea salt & finer spores/pollen and dust particles tend to be concentrated by local meteorological conditions known as inversions (often occured over the L.A. basin in the past).

With cloudy, misty & foggy conditions, as well as reflections from all surfaces, the scattering is independant of wavelength. It's the reason why snow on the ground as well clouds in the sky will take on the color of the local street lights.

The issue of what we sense is another one. Most lighting in our cities provide light-levels for photopic vision. Only in a dark alley or unlit back-yard are you going to use scotopic vision. So the richer yellow light of sodium will be sensed by our photopic vision more readily (peak sensitivity for photopic ≈ 555nm). But if you live in the suburbs and your backyard is completely unlit, after 20 minutes of dark adaptation, you will be using your scotopic vision (where the peak sensitivy ≈ 507nm). When all lights have been converted to LED, the bluer light, scattered easily in a clear dry night, will make things brighter. But there's a trough (depressed area) in the spectrum of most if not all phosphor-based InGaN LEDs which is close to the peak of our scotopic vision. There's also green chlorophyls in grass and foliage to consider (absorbs both blue & red, reflects green).

You see, one of the reasons why light-pollution is continued to be ignored is that no-one has put all the complex science down in a book or major article. Something to consider writing..